Academic journal article Journal of Economics and Finance

Taxes, Time Diversification, and Asset Choice at Retirement

Academic journal article Journal of Economics and Finance

Taxes, Time Diversification, and Asset Choice at Retirement

Article excerpt


This study extends existing research by examining the effect of personal income taxes on the expected relative performance of asset classes as viewed from the retirement date. Results suggest that tax status does not affect the basic conclusions of previous time diversification studies. However, the fund's tax status affects the size of withdrawals that can be sustained, the performance of stocks relative to bonds, and the risk of the retirement fund. In general, for a given size fund and after-tax withdrawal proportion, tax-deferred funds have not only a greater expected return, but also greater risk than non-tax-deferred retirement funds. (JEL: G23)


When viewed in a one-period context under a traditional mean-variance framework, common stocks are on average much riskier than bonds. However, proponents of the concept of time diversification (Levy 1978; Reichenstein 1986; Ambachtsheer 1989) argue that over long horizons common stocks as a whole may not be riskier than bonds. Levy points out that over every 25-year period beginning with January 1926 or later the rate of return on the stock market as a whole has exceeded that of corporate bonds, Treasury bonds, or Treasury bills. Similarly, Siegel (1992) reports that, for every 50-year period since 1872, common stocks as a whole have returned more than bonds.

Butler and Domian (1991, 1992) point out that these findings may be biased in favor of common stocks, since the 25- or 50-year periods overlapped and therefore were not independent. For example, the 25-year periods examined were January 1926 through December 1950, February 1926 through January 1951, and so on. To overcome this weakness, they use empirical resampling to generate independent series of returns. Although their results vary somewhat depending on the subperiod used in their resampling, the benefits of time diversification are apparent, especially over long investment horizons.

Earlier time diversification research focused on the building of wealth, such as building a retirement fund. However, it is also important to determine how the fund should be invested or how to best maintain the fund, on or after retirement. Bengen (1994), Cooley, Hubbard, and Walz (1998), McCabe (1999), and Marbach (2002) focus on the sustainability of withdrawals over the long run. Ho, Milevsky, and Robinson (1994) and Howe (1999) focus on the long-run risk and return properties of portfolios from which one makes periodic withdrawals.

This study extends Howe (1999) by (1) examining more withdrawal proportions in the range retirees are likely to face, especially if their retirement funds are tax-deferred, and (2) examining the effect of personal income taxes on the expected performance of asset classes relative to each other as viewed from the retirement date. It considers the tax implications of either deferring taxes on the fund until the time of withdrawal or paying taxes as the income on the fund is earned. These two scenarios parallel withdrawals from a 401(k), where taxes were deferred, and a situation where the contributions into the fund were not deductible and the income was not tax-deferred. These two situations are not the only conceivable tax implications, but they represent the two extremes.

Survey of Relevant Literature

Studies that support time diversification as a way of reducing risk are of four basic types. The first type involves examining actual wealth relatives over multi-year periods. Examples of these studies are Levy (1978), Siegel (1992), and Ambachtsheer (1989). A weakness of studies such as these is that, to have a large sample of multi-year periods, the periods must overlap. As a result, the periods are not independent, and the findings in favor of stocks may be overstated.

A second type of study is a Monte Carlo simulation study in which the return distributions (along with the autocorrelations and cross-correlations) of the various asset classes are assumed. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.